750 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y Vol. 43, No. 3

E ngFnyring

process development

(17) Grummitt, Oliver, Sensel, E. E., Smit,h, W. R., Burk, R. E., and Lankelma, H. P., I b i d . , 67, 910 (1945).

(18) Hepp, H. J. (to Phillips Petroleum Co.), U. S. Patent 2,461,545 (1949).

(19) Hoskins, V. M., and Ferris, C. A,, IND. ERG. CHEM., AKAL. ED., 8, 6 (1936).

(20) Hughes, E. C., et al., Division of Chemistry, 117th Meeting, -kx. CHEM. Soc., Houston, March 1950.

(21) Ipatieff, V. N., and Grosse, A. V., IND. EXG. CHEM., 28, 461 (1936).

(22) Xomarewsky, V. I., and Click, S. C. , J . Am. Chem. Soc., 69, 492 (1947).

(23) Lien, A. P., and Evering, B. L. [to Standard Oil Co. (Indiana)], U. 8. Patent 2,427,865 (1947).

(24) McCauley, D. 9., Shoemaker, B. H., and Lien, A. P., Division of Petroleum Chemistry, 117th Meeting, AM. CHEY. Soc., Houston, March 1950.

(25) Montgomery, C. W., Mcrlteer, J. H., and Franke, N. W., J . Am. Chem. SOC., 59, 1768 (1937).

Separation of Sulfur and Aromatics from Petroleum

(26) Oblad, G., and Gorin, M. H., IND. ENG. C H E X . , 38, 822

(27 ) Pines, Herman, and Wackher, R. C., J . Am. Chem. Soc . , 68 , 595

(28 ) Podbielniak, W. J., IND. E N G . CHEM., ASAL. E D . , 5, 172 (1933). (29) Powell, T. M., and Reid, E. B., J . Am. Ckem. Soc., 67, 1020

(30) Richmond, J. L. (to Phillips Petroleum Co . ) . U. S. Patent

(1946).

(1946).

(1945).

2,461,568 (1949). (31) Stewart, T. D., and Calkins, W. H., Division of Physical and

Inorganic Chemistry, 115th Meeting, AM. CHEM. SOC., San Francisco, March 1949.

(32) Thomas. C. A,, "Anhydrous Aluminum Chloride in Organic Chemistry," Am. Chem. Soc. Monograph No. 87, S e w York. Reinhold Publishing Corp., 1941.

(33) Whitmore, F. C., Chen,. Eng. ,Vews, 26, 668-74 (1948). (34) Wilcox, L. V., IND. EXG. CHEM., ASAL. ED., 4, 38 (1932).

RECEIVED hIay 15, 1950. Presented before the Division of Petroleum Chemistry a t the 117th Meeting of the AMERICAN CHEMICAL SOCISTI-, Houston, Tex.

WITH BORON FLUORIDE AND HYDROGEN FLUORIDE I

E. C. HUGHES, W. E. SCOVILL, C. H. WHITACRE, R. B. FARIS, J. D. BARTLESON, AND S. M. DARLING THE STANDARD OIL CO. (OHIO), CLEVELAND, OHIO

T h e high-sulfur crudes produced in this country have less value than the sweet crude oils. The sulfur com- pounds are troublesome to refine, requiring special non- corrosible equipment. The presence of the sulfur in prod- ucts is undesirable in every case. The work was under- taken in an effort to provide basic information on a novel method of separating sulfur compounds.

I t was shown that the majority of sulfur compounds in crude oil and even in the heaviest 50% of the crude oil can be removed by the new combination of reagents, and that it is also possible to separate the aromatics.

The significance of these results is twofold. The funda-

OR some time aluminum halides with the halide acid have F been known to form complexes with aromatics ( l a , 13, 15) a,nd aromatics have been separated commercially from lubricating oil fractions with these reagents ( 8 , l l ) . Recently hydrogen fluo-. ride has been described as a reagent for removing sulfur com- pounds (5, 9, 14). Boron fluoride has been used with hydrogen fluoride for treating purposes (3 , 4) . The results of treating a number of stocks with hydrogen fluoride and boron fluoride are given in this paper. Aromatics and sulfur-containing compounds were substantially removed by the solvent system.

APPARATUS AND PROCEDURE The apparat'us and procedures employed are described by

Hughes and Darling ( 7 ) . The boron fluoride and hydrogen fluoride were obtained in cylinders from the Harshaw Chemical Co. and were used as reeeived. Under t,he conditions used in this work, the solvent system consisted of a liquid hydrogen fluoride phase and a gaseous boron fluoride phase.

The apparatus employed was a >Ionel-lined pressure bomb with a 3-inch inside diameter and a capacity of 2.8 liters. The bomb was equipped with three turbinelike stirrers which rot,ated inside a peripheral group of stationary hlonel baffles.

Before each experiment the apparatus was thoroughly dried and flushed with t,ank nitrogen. The hydrogen fluoride and hy- drocarbon charges were forced into the reactor from pressure

mental science shows that the system hydrogen fluoride- boron trifluoride will form complexes with aromatics and sulfur compounds in the same way as Friedel-Crafts cata- lysts, and that it is possible to disassociate these com- plexes at higher temperatures with recovery of the re- agents. Practically it would be possible to use these re- agents for large scale separations of sulfur or aromatic compounds. The process is in competition with solvent extraction processes. The reagent is highly selective and easily recoverable. Preparations of high quality lcerosenes, lubricating oils, and crude oils from stocks of poor quality are suggested as possible uses for the reagent.

vessels which could be weighed to measure the charges. The desired partial pressure of boron fluoride was maintained from a small cylinder, which also could be weighed t,o measure boron fluoride consumption. The react'or was immersed in a water bath. The contents of the reactor were removed through a long siphon tube. The solvent and hydrocarbon phases were usually collected in separate pressure vessels in which they could be weighed. Use of an external connection of flexible saran tubing made it possible to detect the interface between the dark solvent phase and the hydrocarbon phase as the reactor was emptied. The solvent phase was neutralized in iced sodium hydroxide solution and released hydrocarbon was recovered. The hydro- carbon phase was neubralized prior to analysis.

Standard (1, 3 ) test procedures Rere employed in determining the sulfur and aromatic content of the oils. Fluorine was detcr- mined by a method described by Hoskins and Ferris (6). Prior to titration of the fluorine, however, the oil samples were oxidized in a bomb as outlined in the procedure for dekrmination of sulfur in petroleum products ( 1 ) .

Cracking of the hydrocarbon was avoided by controlling tem- perature and quant'ity of the reagents; polymerization, by using nonolefinic stocks. The only variable investigated was the effect of relative quantities of hydrogen fluoride. As the original plan was t o use an excess of hydrogen fluoride, hydrogen fluoride vol- umes from 0.5 t,o 1 part of the hydrocarbon volume were first employed. It was observed that the formation of an interface layer could be avoided and the experimental work simplified by reducing this volume t'o 10% or less of the hydrocarbon volume,

March 1951 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y 75 1

g-O* 80

TABLE I. TREATMENT O F LIGHT GAS OIL FROhI A MISSISSIPPI CRUDE

ExDeriment No. 1 2 3 4

Untreated Raffinate Stock Operating Conditions

Temperature, F. Contact time, min.

0

80 80 82 90 30 30 30 30

8 150 30 60 System pressure, Ib./ss.

Reagent composition based

inch

on hydrocarbon HF, vol. 70 BFaa, wt. 70

Properties Gravity, A.P.I. a t 60' F. Aromatics, vol. % 18 Sulfur, wt. 7' 0,206

39 .7

51 0 51 98.5 0 21 19.4 18 .8

41.0 39.5 44.2 44.9 18 17 6 4 0.116 . , . 0 .071 0.063

Distillation data Initial b.p., F. 424 404 405 352 179 00% b.p., O F. 506 513 511 511 506 Final b.p., F. 580 636 635 620 580 Distillate, % 90 98 98 98 90

r) Raffinate, wt. % 95.6 94.7 82.2 74.1

Properties Aromatics in Engler dis-

Sulfur, wt. 7' tillate, vol. % CONDITIONS:

TEMP."F: = 90' CONTACT TIME = 15 MIN.

BF, PRESSURE= l5Op.s.i,(PARTIAL) Extract

28 . , . 71 84

Distillation data Initial b.p., O F. 50'% b.p., F. Final b.p., O F. Distillate, % Extract, wt. %

420 . . . 370 348 517 . . . . . . 586 , . , 630" 620 87 . . . 97 90

3.7 1 . 5 17.8 21.3

a Includes BFI in vapor phase of autoclave.

so smaller amounts of the reagent were used in subsequent ex- periments.

TREATMENT OF DISTILLATES AND CRUDE OILS

The results obtained by treating a variety of stocks with hy- drogen fluoride-boron fluoride are shown in Tables I and 11. Experiments using hydrogen fluoride and boron fluoride sepa- rately showed that partial desulfurization was effected with hydro- gen fluoride alone, but aromatic removal was not marked.

Substantial improvement in desulfurization was observed through the use of hydrogen fluoride with boron fluoride. The major difference, however, was the removal of most of the aro- matic portions from all the fractions. The hydrocarbon recovered

J I I I I I 0 I O 20 30 40 50

VOLUME PERCENT HYDROGEN FLUORIDE

Treatment of Kerosene with Hydrogen Fluoride and Boron Fluoride

Figure 1.

from the "extract" complex was found to contain 71 to 97% aro- matic compounds. The selectivity of the process for aromatics is thus shown to be high. All of the aromatics were not removed, as shown by residual values of 2 to 6%. This reflected other ob- servations that benzene or toluene in heptane was only partially

TABLE 11. TREATMENT OF SEVERAL DISTILLATES FROM AN ILLINOIS CRUDE Experiment No. 5 6 7

300 Neutral Oil Heavy Naphtha (Dewaxed) Cylinder Stock

Untreated Untreated Untreated stock Raffinate stock Raffinate stock Raffinate

- 4 30 20

110 15

198

150 30

380

49 99.2 12.2 21.7

10 8 . 1

Propertiesb Gravity A.P.I. at 60' F. Aromatics, vol. % Viscosity-gravity constant Viscosity index Sulfur wt. % Raffinkte, wt. %

Properties Aromatics in Engler distillate, vol. Yo Sulfur, wt. 9%

54.0 55 .9 24.3 13 2 . . , . . . 0.853 . . . 58.7

85.5 0 ,053 . . . 0.46

30.5

94.2

56.3

0:Sls 0.10

19.2

6:857 . . . 0.68

27.1

0.800 93.7

43.8 0.12

Extract 97

1 .85 Distillation deta

296 347 423

85 33 .1 7 . 9

55.3

0 Includes BFs in vapor phase of autoclave. b Cylinder stock raffinate dewaxed.

152 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y Vol. 43, No. 3

TABLE 111. EFFECT OB VOLUME OF HYDROGEN FLUORIDE ON REFINING OF KEROSENE"

Operating conditions. BFs pressure. 150 Ib./sq. inch (partial). Contact time 15 minntes. Ternperatire 90' F.

TABLE ITT. Experiment No.

TREATYEKT OF YAZOO CRUDE REDCCED TO 225" F. 17 18 19

Un- Un- treated Raffi- Raffi- treated RafE- Stocka nate nate Stock nste Properties of Raffinate

Expt. Vol. Raffinate, BFs, Extract, A.P.I. bromatics, No. yo Wt. Wt. yo Wt. 7, a t 60' F. Vol. Yo

HF, Gravity Operating conditions Temperature, F . Contact time, mln.

110 110 1 3 15

110 15

50 8 5 .65 89.0 8 . 7 4 . 5 45 .1 8 9 10 88.0 9 . 9 10.3 46.0 4

10 10 86 .2 9 .3 8 . 2 45.4 5 11 15 .6 86.9 12.3 11 .5 46.4 4 12 15.2 87 .7 11.2 11.2 46.4 2 13 20 82.2 15 .3 18 .2 46.2 7 14 30.5 83 .6 14.9 13 .1 47.8 1 16 28.8 84.8 14.9 13 .3 47.2 1.5 16 40 78 .8 17 .5 1 3 . 0 47 .5 3

System pressure. ib./ sq. inch 33 60

Reagent comDosition -based onhydrocarbon

100.2 100.7 HF , vol. % BFs, wt. % 0 14.0

96.5 8 .7

Properties Gravity, h.P.1. a t

Aromatics in Engler 60° F. 29 9 30.7 37 .5 30.2 33.2

distillate, voi. % 54 38 31 36 28 Charge stock 43.3 16

a From Illinois crude having boiling range of 361' to 507' F. and 0.042y0 Sulfur content of this kerosene was reduced to 0.01.54R in experi- sulfur.

ment 11. Viscosity-gravity oon-

Viscosity a t looo F.,

Sulfur, wt. yo 0 . 8 5 0 .65 0.11 0.886 0.32

stant. 0.836 0.826 0.797 0.836 0.818

S.U.S. 8 7 . 1 65.6 82 .7 78 .3

Distillation data Initial b.p., F. 295 360 313 290 340 50% b.p., ",F. 643 650 629 615 664

675 6Q4 Final b.p. , F. TO4 687 667 Distillate, 7G 95 92 96 93 90

separated by the reagents under these conditions. It, therefore, appears that some aromatics do not form as strong complexes as others (IO).

The selectivity was equally good for naphthas and for the heaviest lubricating oil fractions. The production of 90 t o 95 viscosity index oil from Illinois crude lubricating fractions sug- gests the possible utilization of the process in place of solvent extraction. B complex of reagent with aromatics which was in- soluble in both phases formed a t the interface.

This complex formation led to the study of the effect of hydro- gen fluoride volume shown in Table 111. Hydrogen fluoride vol- umes of 10 to 28% of the hydrocarbon phase were found to be ample for removing aromatics from kerosene (Figure 1).

Results of the treatment of some higher sulfur crudes are shown in Tables IV, V, and VI. These crudes are termed "reduced" crudes because the gasoline boiling to 250" F. had been removed before they were treated. With the use of hydrogen fluoride alone, the sulfur removed wad rather meager, but a very good reduction, from 0.85 to 0.11% (experiment 18), was obtained when the combination reagent was used. However, with the use of about half as much boron fluoride (experiment 19), the concentration of sulfur was reduced from 0.87 to only 0.32% Thus the amount of sulfur compounds removed can be

CRUDE OILS.

Raffinate, rvt. yo Interface, wt. Yo

91.2 55.2 69.2 4 .7

Distillation data Initial b.p., F. 507, b.p., F. Final b.p., F. Distillate, ?& Extract, wt. Interface, wt. %

466 252 322 713 726 74 15 .4 10.8

590 619 63 42 10.1 34.7

a 7.157, distillate removed. b Extract of experiment 19 heated to 350" F.

controlled hy thc quantity of boron fluoride provided. Reduced West Tevas crude containing 2.26% sulfur was separated into a 55.7% raffinate having only 0.30y0 sulfur.

A comparison of the distribution of the products of the crude and the raffinate (Table VII) shows several major differences be- tween thc crude oil and the raffinate. The aromatic content of the distilled fractions was very much lower for the rafinate. In particular, the material boiling above 1000 F. (atmospheric

T.&BLE 1'. TREATMENT O F WIGST TEXAS CRUDE REDUCED TO 250' F.

Experiment No. 20 21 (Vapor

Untreated Phase) Stock Raffinate Raffinate

Operating conditions Temperature, F. 100 100 Contact time min. 15 60 System presske, lb./sq. inch 165 Stmospheric

Reagent composition based on hydro- carbon

HF, vol. % 10 3 . 5 BFa, wt. % 8 . 8 11.2

PropertiesQ Gravity, A.P.1. at 60' F. 25 .7 37 .8 35.0 Aromatics in Engler distillate,

Viscosity-gravity constant

Sulfur, wt. % 2.26 0.30 0 .55 Fluorine, wt. yo 0.058 0 .06 Raffinate, wt. % 55.7 62.9

0 .'866 0:817 Q:g27 vel. ,% Viscosity a t 100' F., S.U.S. 77.2 43 .6 . . .

(Vapor Phase)

Untreated Raffinate, Stock" Expt. 22

Operating condit;oih Tempciature, F. Contact time. min. System pressure. ib./sq. inch

110 30

Atmospheric Reagent composition based on hydrocarbon

HF, voliime '3, BPI, wt. %

I . 8 5 . 0

30 .1 Propertiesb

Gravity, A.P.I. a t 60' F. Aromaties in Eneler distillate. vol. To

34.4

0.847

1 .74 55 ,5

0 826

0 43 0 01

49 1

78 2

. - Viscosity-gravity constant Viscosity a t 100' F., S.U.S. Sulfur, wt. % Fluorine, wt. ?& Raffinate, wt.

. . .

Propertiesb

""I. al, Aromatics in Engler distillate,

PropertiesC Extract Aromatics in Engler distillate, vol. % Sulfur, wt. yo 4 . 4 1 Fluorine, wt. % 0.07d Extract, wt. yo 21.3

. . . Extract

5.76 5.12 0.02e 0.06

40.6 37 .1

SUI<& 'Gt. % Fluorihe, wt. % Extract, wt. yo Interface, wt. % 12.9% distillate removed.

b Raffinate heated to 350' F. C Extract heated to 500' F. d Fluorine determined on extract heated to 600-850° I:. in 1Iaatelloy col-

umn.

0 Raffinate heated to 350' F. b Extracts heated to 550° F. in Hastelloy column. C Fluorine determined on extract heated to 600-650O F.

March 1951 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y 753

TABLE VII. COMPARATIVE DISTILLATION OF WEST TEXAS CRUDE OIL (250” F. REDUCED) AXD HYDROGEN FLUORIDE-BORON FLUORIDE RAFFINATE

(Experiment 20 of Table V) Fraction.

A.P.I. Temp.

to Atmos. Pressure, Wcight % % Sulfur Gravity Viscosity at 100’ % iiromatics Corrected

Press., F. Rlm. Crude Raff.’ Crude Raff. Crude Raff. Crude Raff. * Crude Raff. <400 400-550 550-700 700-835 835-1005 Over 1005

Atmos. 10 10

0 .2 0 . 2

. . .

18 1 22.3 18 .5 23 .8 17 .7 22 5 5 2 6 . 4 7 8 7 .6

32.0 16 .2

0 .46 0.01 1 .27 0.04 2 .10 0.17 2 .34 0.18 2 .65 0.31 3.99 1 .00

46 .7 54 .3 36.2 46 .1 27.6 39 .9 23.7 35 .5 20 .5 34.7 . . 29.8

0.92CS 2.36CS 9 . 0 2 c s

29.54CS Waxy Bs1)haItic

0.97CS 2 . 3 3 c s 8.28CS

23.94CS Waxy Waxy

29 19 32 17 41 19 49 . . . . . .

pressure) was segregated into the “extract.” This left the raffi- nate a very high-gravity, low-sulfur crude oil.

The results obtained with these reagents seemed better than necessary for most refinery use, and as a result the use of hydrogen fluoride gas instead of the liquid was investigated. The hydrogen

crude oil and the whole was agitated at atmospheric pressure. The amount of hydrogen fluoride utilized was only 3.5y0 of hy- drocarbon phase, and this, along with the 11.2 weight yo of boron fluoride taken up, separated out 37.1% of the crude. The 63% remaining as raffLnate was a 35 gravity, o.55y0 sulfur oil and, therefore, a satisfactory sweet crude oil. Similar treatment of Midway crude removed 21% of the crude, leaving a 34.4 gravity product containing 0.43% sulfur.

1. fluoride was mixed with the gaseous boron fluoride above the

REAGENT RECOVERY

In much of the work the hydrocarbon phases were recovered by washing with alkali. However, the advantage of the hydrogen fluoride-boron fluoride system over other halides lies in the ability to distill them from the final products. Although no difference was found in the ra fha tes whether alkali or distillation was used to remove the reagents, small changes took place in some of the “extracts.” The distillation was carried out in a 1-liter Monel

8

~~

TABLE VIII. COMPARISON O F “EXTRACT” PROD7JCT AFTER REMOVAL OF REAGENTS BY HYDROLYSIS AND BY DISTILLATION

(Experiment 19 of Table IV) Stock. Conditions. Vol. % H F 96.5. Wt. % BFa 6.7. Temperature, F. l l O o .

Yazoo crude (reduced to 225O F.)

Engler distillation, O F. Initial b.p.

i n s

Residie, 7’ Sulfur. wt. d Asphaltenes;-wt. % Optical density Fluorine wt. % Fluorine’on residus. wt. .%

tr dis- Aromatics in Engle tillate, vol. %

A@

310 560 598 618 633 672 670 30

18,300

3.46 2.97

... . . * 91

Bb 322 570 636 675 701 713 724 23

11 .4 39,200

3 .40

0.47 0.14

a2 * 0 Reagent neutralized with NHa then washed with water.

b Reagent distilled from extraht a t maximum liquid temperature of 350‘ F.

flask containing a thermometer well immersed in the liquid. No column was used. The maximum liquid temperature reached mas 350’ F. Mild polymerization was an evident reaction during distillation of the reagent from the extract hydrocarbon, as shown by examples in Table VIII. The fluorine content left in the ex- tract product by this procedure was 0.47%. However, on Engler distillation the residue in the flask contained only o.14y0 fluorine, indicating that higher temperatures under reflux could remove more fluorine for possible re-use.

In order to evaluate how much fluorine would remain in the

extract a t higher reflux temperatures, a small column was con- structed of Hastelloy B. In this column the extract was gradu- ally raised to a temperature of 550” to 650” F. Reflux was established in the column and the fluorine compounds were removed but not identified. except to note that they were acidic in nature. The residual fluorine in several extracts shown in Tables V and VI was reduced to 0.02 and 0.07% by this opera- tion.

SUMMARY

Substantial desulfurization and removal of aromatics from a number of distillates and crudes were observed with the use of hydrogen fluoride and boron trifluoride. The reagent has the advantages of high selectivity and easy recovery. Selectivity is shown to be equally as good for naphthas as for the heaviest lubricating oils. The use of hydrogen fluoride as a gas instead of the liquid in admixture with boron trifluoride was also in- vestigated, and satisfactory results were obtained on treating relatively high-sulfur crudes.

ACKNOWLEDGMENT

The authors wish t o acknowledge the assistance given them by J. J. Szabo in the experimental phase of this work, by E. R. Kosman in the preparation of the manuscript, and by R. E. Burk in the direction of the research. The authors also wish to express their appreciation to The Standard Oil Co. (Ohio) for the release of the information.

LITERATURE CITED

Am. Soc. Testing Materials, A.S.T.M. designation D 129-44.

Burk, R. E. [to Standard Oil Co. (Ohio)], U. S. Patent 2,343,744 (1944).

Ibid., 2,343,841 (1944). Foster, A. L., Oil Gas J . , 45, No. 6, 96 (June 15, 1946). Hoskins. W-. M.. and Ferris. C. A,. IND. ENQ. CHEM.. ANAL.

Ibid., D 875-46T.

ED., 8, 6-9 (1936). Hughes, E. C.. and Darling, S. M., IND. ENG. CHEM., 43,

726 (1951). Kalichevsky, V. A., “Modern Methods of Refining Lubricating

Oils,’’ AM. CHEM. Soo. Monograph 76, p. 91, New York, Rein- hold Publishing Corp., 1938.

Lien, A. P., McCaulay, D. A., and Evering, B. L., IND. ENG. CHEM., 41,2698 (1949).

McCaulay, C. A., Shoemaker, B. H., and Lien, A. P., Division of Physical and Inorganic Chemistry, 117th Meeting, AM. CHEM. SOC., Detroit, 1950.

Musselman. J. M. [to Standard Oil Co. (Ohioll. U. S. Patent

-

. . _,

2,020.954 (1935). . Norris, J. F., and Innraham, J. N., J . Am. Chem. Soc., 62,

1298 (1940). Norris, J. F.. and Wood, J. E., 111, Ibid., 62, 1428-32 (1940).

(14) Scafe, E. T., Petroleum Refiner, 25, No. 9, 87 (1946). (15) Thomas, C. b., “Anhydrous Aluminum Chloride in Organic

Chemistry,” AM. CHEM. SOC. Monograph 87, p. 56, New York , Reinhold Publishing Corp., 1941.

RECEIVED May 15, 1950. Presented before the Division of Petroleum Chemistry a t the 117th Meeting of the AMERICAN CHEMICAL SOCIETY, Houston. Tex.